Hi Anton, one more thing about the destroy. This seems to be an issue of the simulator plugin. The ipmidirect gets completely "destroyed" as far as I see. Cheers, Uli
> -----Original Message----- > From: Kleber, Ulrich (NSN - DE/Munich) > Sent: Monday, August 16, 2010 10:13 AM > To: 'ext [email protected]' > Cc: [email protected] > Subject: RE: oHpi API > > Hi Anton, > This is great! > I used your patch for marshall and commented line 2856 in openhpid. > > Then I started a daemon with just a simulator plugin and created > dynamically a new ipmidirect plugin. It worked. > And it is really loaded. The hpitest PASSed all the testcases that > passed with the statically loaded plugin. > > Should I contribute the ohcontrol client application and write a > bug report for destroy? > > Maybe we should even do that preliminary change to openhpid.cpp. I > guess there is some dirty programming causing the hash table still > be accessed after the HandlerCreate has completed. I could add a > comment to that line and write another bug report. > > Cheers, > Uli > > > -----Original Message----- > > From: ext [email protected] > > [mailto:[email protected]] > > Sent: Friday, August 13, 2010 6:32 PM > > To: Kleber, Ulrich (NSN - DE/Munich) > > Cc: ext Anton Pak; [email protected] > > Subject: Re: oHpi API > > > > Uli, > > > > o) list, getnext, info worked fine for me now. It was my > > configuration issue. > > > > o) In rev 7133 I fixed marshalling for oHpiParamGet and > oHpiParamSet. > > Could you try it? > > > > o) There are a typo in ohcontrol create help: > > - libsimulater -> libsimulator > > > > o) When I comment line 2856 "g_hash_table_destroy(config_table);" in > > openhpid/openhpid.cpp, ohcontrol create works fine. > > > > o) seems ohcontrol destroy does something. After destroy, list > > shows no handler with destroyed id. However the resources for > > this handler are still present and sensor reading on these resources > > works ok. Seems the handler was not destroyed completely. > > > > Anton Pak > > > > > > > Hi Anton, > > > as we discussed yesterday in the chat room, here I send you > > the client > > > program that I created to work with the oHPI APIs. > > > The attached archive contains the client and the patch for > > the Makefile > > > for your convenience. > > > > > > I did only very few tests yet, but found a few issues > with the APIs. > > > - The APIs work only with the default domain. > > > - oHpiHandlerInfo, oHpiHandlerGetNext, oHpiHandlerDestroy seem ok. > > > - oHpiHandlerFind always returns invalid session. > > > This function is the only one that already has a > session parameter > > > (which is needed for other domains) so I think the introduction > > > of that parameter was not completed at that time. > > > - oHpiHandlerRetry I didn't test yet. I think it only works when a > > > handler is defined completely but couldn't initialize or > > was terminated > > > for some other reason than oHpiHandlerDestroy. > > > - oHpiHandlerCreate I couldn't verify yet. It always > > returns some error. > > > Since you were able to call it from within the daemon, > I think the > > > parameters are not copied through. > > > - oHpiParamGet and oHpiParamSet for OHPI_DAT_SAVE always > > runs into the > > > following error: > > > ohcontrol: marshal.c:468: Marshal: Assertion `0' failed. > > > Aborted > > > other parameters seem to work. > > > > > > What say? > > > > > > Cheers, > > > Uli > > > <<u7130-6.tgz>> > > > > > > > > > > > > > > > > > > > > > ------------------------------------------------------------------------------ This SF.net email is sponsored by Make an app they can't live without Enter the BlackBerry Developer Challenge http://p.sf.net/sfu/RIM-dev2dev _______________________________________________ Openhpi-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/openhpi-devel
